Upload 9 files

app.py

@@ -119,16 +119,12 @@ def generate():
         sys.stdout.flush()
         generation_start = time.time()
         
-        # Clamp temperature to avoid numerical issues
-        temperature = max(0.5, min(temperature, 1.5))
-        
+        # Use greedy decoding to avoid temperature/sampling issues
         with torch.no_grad():
             outputs = model.generate(
                 **inputs,
                 max_new_tokens=max_new_tokens,
-                temperature=temperature,
-                top_p=top_p,
-                do_sample=True,
+                do_sample=False,  # Use greedy decoding instead of sampling
                 pad_token_id=tokenizer.eos_token_id,
                 eos_token_id=tokenizer.eos_token_id,
                 repetition_penalty=1.1,